Virtual horizontal stereo camera

Abstract

An apparatus including a stereo camera and a processor. The stereo camera may comprise a first capture device and a second capture device in a vertical orientation. The first capture device may be configured to generate first pixel data and the second capture device may be configured to generate second pixel data. The processor may be configured to receive the first pixel data and the second pixel data, generate a vertical disparity image in response to the first pixel data and the second pixel data, generate a virtual horizontal disparity image in response to the first pixel data and the vertical disparity image and detect objects by analyzing the vertical disparity image and the virtual horizontal disparity image. An analysis of the virtual horizontal disparity image may enable the processor to detect the objects not detected in the vertical disparity image alone.

The invention claimed is: 
     
       1. An apparatus comprising:
 a first capture device configured to generate first pixel data; 
 a second capture device configured to generate second pixel data, wherein said first capture device and said second capture device are implemented in a vertical orientation; and 
 a processor configured to (i) process said first pixel data and said second pixel data arranged as a vertical disparity image, (ii) generate a virtual horizontal disparity image in response to (a) said first pixel data and (b) said vertical disparity image and (iii) detect objects by analyzing (a) said vertical disparity image and (b) said virtual horizontal disparity image, wherein an analysis of said virtual horizontal disparity image enables said processor to detect said objects not detected in said vertical disparity image alone. 
 
     
     
       2. The apparatus according to  claim 1 , wherein said vertical orientation comprises said first capture device located above said second capture device. 
     
     
       3. The apparatus according to  claim 1 , wherein said virtual horizontal disparity image comprises an intelligent estimation performed by said processor of a horizontal disparity image that would be generated in response to (a) third pixel data generated from a third capture device in a horizontal orientation with said first capture device and (b) said first pixel data. 
     
     
       4. The apparatus according to  claim 1 , wherein (i) said processor is configured to implement a convolutional neural network and (ii) said convolutional neural network is configured to generate said virtual horizontal disparity image. 
     
     
       5. The apparatus according to  claim 4 , wherein said convolutional neural network is configured to generate said virtual horizontal disparity image in response to providing said first pixel data as input to a directed acyclic graph. 
     
     
       6. The apparatus according to  claim 5 , wherein said directed acyclic graph is generated by said convolutional neural network in response to training data. 
     
     
       7. The apparatus according to  claim 6 , wherein said training data is generated in response to (a) said first pixel data generated by said first capture device, (b) third pixel data from a third capture device and (c) said vertical disparity image. 
     
     
       8. The apparatus according to  claim 7 , wherein (i) said first capture device and said second capture device are implemented as a first stereo camera, (ii) said third capture device is implemented as a component of a second stereo camera, (iii) said second stereo camera comprises said third capture device and a fourth capture device and (iv) said third capture device and said fourth capture device are in said vertical orientation. 
     
     
       9. The apparatus according to  claim 8 , wherein (i) said first stereo camera and said second stereo camera are mounted on a vehicle and (ii) said vehicle is used to enable capturing a large data set of said training data. 
     
     
       10. The apparatus according to  claim 9 , wherein training used for generating said directed acyclic graph is performed offline without said first stereo camera and said second stereo camera attached to said vehicle. 
     
     
       11. The apparatus according to  claim 7 , wherein said third capture device is located in a horizontal orientation with respect to said first capture device. 
     
     
       12. The apparatus according to  claim 1 , wherein said apparatus is implemented as part of a computer vision system for a vehicle to generate an all-around view of said vehicle. 
     
     
       13. The apparatus according to  claim 1 , wherein said apparatus is configured to capture a rear-view of a vehicle. 
     
     
       14. The apparatus according to  claim 1 , wherein said objects not detected in said vertical disparity image alone comprise thin vertically oriented objects that result in a singularity in a disparity calculation. 
     
     
       15. The apparatus according to  claim 14 , wherein said disparity calculation used to generate said vertical disparity image comprises a semi-global matching calculation. 
     
     
       16. The apparatus according to  claim 14 , wherein said thin vertically oriented objects comprise poles. 
     
     
       17. The apparatus according to  claim 1 , wherein said processor generates said virtual horizontal disparity image by filling in missing disparity values. 
     
     
       18. A method for generating a virtual horizontal disparity image, comprising the steps of:
 receiving first top pixel data and first bottom pixel data from a first vertically oriented stereo camera; 
 receiving second top pixel data from a second vertically oriented stereo camera, wherein said first top pixel data is in a horizontal orientation with said second top pixel data; 
 generating training data for a virtual horizontal disparity image in response to said first top pixel data, said first bottom pixel data and said second top pixel data; 
 generating a vertical disparity image in response to said first top pixel data and said first bottom pixel data; and 
 generating a virtual horizontal disparity image in response to (a) said first top pixel data, (b) said vertical disparity image and (c) said training data, wherein
 (a) said training data is generated with said first vertically oriented stereo camera implemented together with said second vertically oriented stereo camera, and 
 (b) said virtual horizontal disparity image is generated with said first vertically oriented stereo camera implemented without said second vertically oriented stereo camera. 
 
 
     
     
       19. The method according to  claim 18 , further comprising the step of:
 detecting objects by analyzing (a) said vertical disparity image and (b) said virtual horizontal disparity image, wherein an analysis of said virtual horizontal disparity image enables detecting said objects using a processor that are not detected by said processor using said vertical disparity image alone. 
 
     
     
       20. An apparatus comprising:
 an interface configured to receive (i) first pixel data and (ii) second pixel data; and 
 a processor configured to (i) connect to said interface, (ii) generate a horizontal disparity image in response to (a) said first pixel data and (b) said second pixel data, (iii) generate a virtual vertical disparity image in response to (a) said first pixel data and (b) said horizontal disparity image and (iv) detect objects by analyzing (a) said horizontal disparity image and (b) said virtual vertical disparity image, wherein
 (a) an analysis of said virtual vertical disparity image enables said processor to detect said objects not detected in said horizontal disparity image alone, 
 (b) said first pixel data and said second pixel data are generated by a stereo camera, 
 (c) said stereo camera comprises a first capture device and a second capture device implemented in a horizontal orientation, 
 (d) said first capture device is configured to generate said first pixel data, and 
 (e) said second capture device is configured to generate said second pixel data.